U.S. patent application number 14/579168 was filed with the patent office on 2015-07-23 for geographic atrophy identification and measurement.
The applicant listed for this patent is Kabushiki Kaisha TOPCON. Invention is credited to Charles A. Reisman, Qi Yang.
Application Number | 20150201829 14/579168 |
Document ID | / |
Family ID | 52354874 |
Filed Date | 2015-07-23 |
United States Patent
Application |
20150201829 |
Kind Code |
A1 |
Yang; Qi ; et al. |
July 23, 2015 |
GEOGRAPHIC ATROPHY IDENTIFICATION AND MEASUREMENT
Abstract
Geographic atrophy of the eye can be detected and measured by
imaging the eye at a depth greater than the retinal pigment
epithelium (RPE) at a plurality of locations of the eye, for
example, using optical coherence tomography (OCT); determining a
ratio of the intensities of imaging signals of a retinal layer(s)
with respect to the intensity of imaging signals of a sub-RPE
layer(s) at each location; determining representative values based
at least in part on the determined ratios; generating a map of the
representative values; and identifying diseased areas from the map.
Contours and binary maps may be generated based on the identified
diseased areas. The size and shape of the identified areas may be
analyzed and monitored over a period of time.
Inventors: |
Yang; Qi; (Foster City,
CA) ; Reisman; Charles A.; (Mamaroneck, NY) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
Kabushiki Kaisha TOPCON |
Tokyo |
|
JP |
|
|
Family ID: |
52354874 |
Appl. No.: |
14/579168 |
Filed: |
December 22, 2014 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
61929743 |
Jan 21, 2014 |
|
|
|
Current U.S.
Class: |
382/131 |
Current CPC
Class: |
G06T 2207/30041
20130101; A61B 3/1025 20130101; G06T 2207/10101 20130101; G06K
2009/4666 20130101; A61B 3/12 20130101; G06K 9/00617 20130101; G06K
9/52 20130101; A61B 3/1225 20130101; G01N 2021/887 20130101; G06K
9/4604 20130101; A61B 5/4842 20130101; G06T 7/0012 20130101; A61B
3/0025 20130101; G06T 5/002 20130101; G01N 2021/1787 20130101; A61B
3/102 20130101; G01N 21/4795 20130101; G01B 9/02091 20130101; G01N
21/6456 20130101; G06K 9/4671 20130101 |
International
Class: |
A61B 3/00 20060101
A61B003/00; G01N 21/64 20060101 G01N021/64; G01N 21/47 20060101
G01N021/47 |
Claims
1. A method of processing acquired ophthalmic image data
comprising: providing an imaging signal produced by an imaging
device capable of penetrating beyond the retinal pigment epithelium
(RPE) of a subject's eye; detecting intensities of the imaging
signal as the imaging signal is backscattered by each of a
plurality of tissue layers in the subject's eye for a plurality of
axial scans; determining a ratio of the intensities of the
backscattered imaging signals, the ratio being the intensity of the
backscattered imaging signal of at least a portion of a retinal
layer with respect to the intensity of the backscattered imaging
signal of at least a portion of a sub-RPE layer for each of the
plurality of axial scans; and determining a representative value
for each of the plurality of axial scans based at least in part on
the determined ratio for the corresponding axial scan.
2. The method of claim 1, wherein said at least a portion of a
retinal layer comprises a combination of retinal layers, and
wherein said at least a portion of a sub-RPE layer comprises a
combination of sub-RPE layers.
3. The method of claim 1, wherein the representative value is
selected from the group consisting of: attenuation coefficients,
integrated attenuation, or a monotonic or near-monotonic proxy
measurement.
4. The method of claim 1, wherein the method further comprises:
identifying diseased areas of the subject's eye based at least in
part on the determined representative values.
5. The method of claim 4, wherein the step of identifying diseased
areas of the subject's eye further comprises: generating a map of
the representative values or ratios; generating seeds of diseased
areas; removing outlier seeds; growing a region encompassed by the
generated seeds that were not removed; refining a contour of the
grown regions; identifying an area inside the contour as diseased;
and outputting the generated map with a contour around the regions
identified as diseased or outputting a binary mask of the regions
identified as diseased.
6. The method of claim 5, wherein the step of generating seeds is
performed by removing noise from the generated map and applying a
thresholding technique on the generated map.
7. The method of claim 6, wherein the thresholding technique
comprises finding an Otsu threshold of the generated map, comparing
the Otsu threshold with a pre-set value, and selecting an intensity
threshold based on the comparison, wherein seeds are generated
using pixels of the map that have intensities lower than the
selected intensity threshold.
8. The method of claim 5, wherein the step of removing outlier
seeds is performed by grouping connected seed components and
applying a distance analysis on the generated seeds.
9. The method of claim 1, wherein the imaging signal has a center
wavelength of at least 1 .mu.m.
10. The method of claim 1, wherein the retinal layer, portion of
the retinal layer, combination of retinal layers, sub-RPE layer,
portion of a sub-RPE layer, or combination of sub-RPE layers is
determined using polarization sensitive optical coherence
tomography (PS-OCT).
11. The method of claim 1, wherein the method is used to determine
which locations of the subject's eyes are diseased; an area of
individual, disease affected regions; a number of individual,
disease affected regions; a total area of disease affected regions;
a circumference of individual disease affected regions, or a total
circumference of disease affected regions.
12. A method of processing acquired ophthalmic image data
comprising: providing an imaging signal produced by an imaging
device capable of penetrating beyond the choroid/sclera interface
of a subject's eye; detecting intensities of the imaging signal as
the imaging signal is backscattered by each of a plurality of
tissue layers in the subject's eye for a plurality of axial scans;
determining a ratio of the intensities of the backscattered imaging
signals for each of the plurality of axial scans, the ratio being
the intensity of a first portion of the backscattered imaging
signal with respect to the intensity of a second portion of the
backscattered imaging; and determining a representative value of
each of the plurality of axial scans based at least in part on the
determined ratio for the corresponding axial scan.
13. The method of claim 12, wherein the representative value is
selected from the group consisting of: attenuation coefficients,
integrated attenuation, or a monotonic or near-monotonic proxy
measurement.
14. The method of claim 12, wherein the method further comprises:
identifying diseased areas of the subject's eye based at least in
part on the determined representative values.
15. The method of claim 14, wherein the step of identifying
diseased areas of the subject's eye further comprises: generating a
map of the representative values or ratios; generating seeds of
diseased areas; removing outlier seeds; growing a region
encompassed by the generated seeds that were not removed; refining
a contour of the grown regions; identifying an area inside the
contour as diseased; and outputting the generated map with a
contour around the regions identified as diseased or outputting a
binary mask of the regions identified as diseased.
16. The method of claim 15, wherein the step of generating seeds is
performed by removing noise from the generated map and applying a
thresholding technique on the generated map.
17. The method of claim 16, wherein the thresholding technique
comprises finding an Otsu threshold of the generated map, comparing
the Otsu threshold with a pre-set value, and selecting an intensity
threshold based on the comparison, wherein seeds are generated
using pixels of the map that have intensities lower than the
selected intensity threshold.
18. The method of claim 15, wherein the step of removing outlier
seeds is performed by grouping connected seed components and
applying a distance analysis on the generated seeds.
19. The method of claim 12, wherein the imaging signal has a center
wavelength of at least 1 .mu.m or is a polarization sensitive
optical coherence tomography (PS-OCT) signal.
20. The method of claim 12, wherein the method is used to determine
which locations of the subject's eyes are diseased; an area of
individual, disease affected regions; a number of individual,
disease affected regions; a total area of disease affected regions;
a circumference of individual disease affected regions, or a total
circumference of disease affected regions.
21. A method of processing acquired ophthalmic image data
comprising: providing an imaging signal produced by a imaging
device, the imaging signal having a center wavelength of at least 1
.mu.m or being a polarization-sensitive optical coherence
tomography (PS-OCT) signal; detecting intensities of the imaging
signal as the imaging signal is backscattered by each of a
plurality of tissue layers in a subject's eye for a plurality of
axial scans; determining a ratio of the intensities of the
backscattered imaging signals for each of the plurality of axial
scans, the ratio being the intensity of a first portion of the
backscattered imaging signal with respect to the intensity of a
second portion of the backscattered imaging; identifying diseased
areas of the subject's eye based at least in part on the determined
ratio.
22. The method of claim 21, further comprising determining a
representative value of each of the plurality of axial scans based
at least in part on the determined ratio for the corresponding
axial scan, wherein the representative value is selected from the
group consisting of: attenuation coefficients, integrated
attenuation, or a monotonic or near-monotonic proxy
measurement.
23. The method of claim 22, wherein the step of identifying
diseased areas of the subject's eye further comprises: generating a
map of the representative values or ratios; generating seeds of
diseased areas; removing outlier seeds; growing a region
encompassed by the generated seeds that were not removed; refining
a contour of the grown regions; identifying an area inside the
contour as diseased; and outputting the generated map with a
contour around the regions identified as diseased or outputting a
binary mask of the regions identified as diseased.
24. The method of claim 23, wherein the step of generating seeds is
performed by removing noise from the generated map and applying a
thresholding technique on the generated map.
25. The method of claim 24, wherein the thresholding technique
comprises finding an Otsu threshold of the generated map, comparing
the Otsu threshold with a pre-set value, and selecting an intensity
threshold based on the comparison, wherein seeds are generated
using pixels of the map that have intensities lower than the
selected intensity threshold.
26. The method of claim 23, wherein the step of removing outlier
seeds is performed by grouping connected seed components and
applying a distance analysis on the generated seeds.
27. The method of claim 21, wherein the method is used to determine
which locations of the subject's eyes are diseased; an area of
individual, disease affected regions; a number of individual,
disease affected regions; a total area of disease affected regions;
a circumference of individual disease affected regions, or a total
circumference of disease affected regions.
Description
CROSS REFERENCE TO RELATED APPLICATIONS
[0001] This application claims priority to U.S. Provisional
Application Ser. No. 61/929,743, filed on Jan. 21, 2014, entitled
"GEOGRAPHIC ATROPHY IDENTIFICATION AND MEASUREMENT", the entirety
of which is incorporated herein by reference.
BACKGROUND OF THE INVENTION
[0002] 1. Field of the Invention
[0003] This application relates generally to the identification and
measurement of geographic atrophy, and more specifically, to the
identification of geographic atrophy in ophthalmic optical
coherence tomography applications.
[0004] 2. Description of Related Art
[0005] In ophthalmic and other applications, optical coherence
tomography (OCT) is frequently used to generate three-dimensional
scan data of a volume. OCT scans typically utilize one of
relatively few fixation positions--most commonly the macula or
optic disc, which place the respective feature at or near the
center of the OCT scan. Wide scan patterns, which cover a large
area that encompasses both the macula and optic disc, are also
increasingly popular. A single scan in the axial dimension
generates a depth profile (an "A-line" or "A-scan"), while a series
of A-scans along a given line generate a B-scan. A series of
B-scans can then be used to form the 3D volume.
[0006] In certain diseases of the back of the eye--such as
geographic atrophy (GA), choroideremia, retinitis pigmentosa,
glaucoma, and multiple sclerosis (MS)--there are rough patterns of
tissue atrophy that may present as either or both of: (1) changes
in thickness; and (2) changes in OCT reflectivity values (image
pixel intensities). Changes in thickness tend to result in thinning
of the affected tissues with disease progression; however, less
common retinal diseases related to inflammation can cause an
increase in thickness. Diseases including GA, choroideremia, and
retinitis pigmentosa involve atrophy of a layer or layers in the
outer retina and/or surrounding tissue layers such as the retinal
pigment epithelium (RPE), outer segments, inner segments, and/or
choriocapillaris. Glaucoma and MS typically involve atrophy of the
retinal nerve fiber layer (rNFL) and ganglion cell layer (GCL).
Changes in OCT reflectivity values typically result from or are
associated with a change in tissue attenuation properties,
sometimes in a different tissue layer than the one being measured
or observed. Alternatively and spuriously, decreases in
reflectivity can result from increased shadowing (darkening,
reducing the dynamic range of data) associated with lens
crystallizations (cataracts) or floaters in the vitreous.
Relatively large measurable changes are more associated with
certain tissue layers and with certain retinal locations than with
others, and the nature of these is highly disease specific.
[0007] Geographic atrophy, also called advanced `dry` age-related
macular degeneration (AMD), causes substantial and progressive
visual impairment, developing in approximately 20% of patients
presenting with preexisting clinical signs of AMD. GA is
characterized by confluent areas of apoptosis at the level of
photoreceptors and RPE atrophy, occurring bilaterally in more than
half of patients affected. The condition progresses slowly over
time, typically sparing the fovea until the later stages of disease
progression. The atrophy can be unifocal (one atrophic spot) or
multifocal (multiple spots).
[0008] Clinical trials to evaluate new therapies for
non-neovascular AMD require reliable, accurate, and simple means of
monitoring GA size and progression. Accurately monitoring GA
progression can also help to better understand the pathogenesis of
GA and AMD in general, as numerous aspects of AMD pathogenesis are
not particularly well understood at present.
[0009] In choroideremia, the choriocapillaris (small capillary
vessels in the choroid just outer to the Bruch's membrane), the
RPE, and photoreceptors (in the later stages of disease) degenerate
leading to lost visual function over time. As with GA, a visible
thinning of the RPE can often be observed in affected areas in OCT
scans. Additionally, due to decreased attenuation of the RPE layer
(primarily of the RPE complex, though also of the choriocapillaris
and possibly other tissue features, such as the photoreceptors),
the signal in the choroid and beyond (e.g., the sclera) appears
relatively bright in OCT scans.
[0010] Retinitis pigmentosa is a progressive retinal disease that
affects the photoreceptors resulting in a severe loss of vision.
Atrophy can be observed in the outer segments (OS) of receptor and
other layers, such as the outer nuclear layer (ONL). Usually the
thinning of the OS layer precedes changes in other receptor layers.
In the case of RP, the visible thinning of the OS, possibly the
RPE, ONL, and total retinal thickness can be observed with the NFL
layer intact or even thicker.
[0011] There are four main processes in age-related macular
degeneration pathogenesis, which preferentially affects the macula.
In the first, lipofuscin formation, RPE metabolic insufficiency
associated with aging leads to progressive accumulation of
lipofuscin granules (a roughly even mixture of lipids and proteins)
in the RPE. This is also related to failure to clear some
metabolites from outer segment phagocytosis from the RPE. A
lipofuscin component known as A2E is known to be a cytotoxic
molecule, capable of generating free radicals, damaging DNA,
etc.
[0012] Next, drusen formation is the result of extracellular
deposits collecting between the RPE and Bruch's membrane. While
most elderly individuals have a small number of "hard" drusen, the
presence of numerous "hard" or "soft" drusen (especially the soft
variety, which are typically larger in area), particularly when
accompanied by pigment changes, is thought to be an early indicator
of AMD. Drusen formation is also thought to relate to inflammatory
processes as well as CFH gene allele Y402H.
[0013] The third process, chronic local inflammation, is not very
well understood and is tied to lipofuscin and drusen formation, as
well as additional factors including light irradiation and
genetics, including the CFH gene allele Y402H,
[0014] Finally, neovascularization (wet AMD) is distinct from
geographic atrophy. Neovascularization is thought to be preceded
either by hypoxia or inflammation (or a combination of the two),
leading to a signaling pathway that triggers an increased
production of VEGF (vascular endothelial growth factor), which
precipitates choroidal neovascularization.
[0015] Geographic atrophy is generally considered to be the non-wet
end-stage of AMD, although some consider GA to be the default
end-stage of advanced AMD. It should be noted, however, that both
GA and wet AMD can occur together in the same eye. In GA, the RPE
and outer segments atrophy in affected regions, and the atrophy
typically extends to surrounding tissue layers as well, including
the inner segments, the outer nuclear layer, and possibly the
choriocapillaris. Associated with the loss of retinal function,
blind spots (scotomas) result in the patient's central vision.
[0016] Traditional imaging modalities, fundus imaging and fundus
autofluorescence imaging, have been used to detect GA. In fundus
imaging, GA is defined as a sharply demarcated area exhibiting an
apparent absence of the RPE, with visible choroidal vessels and no
neovascular AMD. Fundus autofluorescence imaging is based on the
autofluorescence properties of AMD-related compounds, such as
lipofuscin, that build up in RPE cells. Fundus autofluorescence
imaging is probably the most widely applied technique with respect
to GA detection at present.
[0017] In an emerging OCT technique, GA is associated with
increased OCT signal intensities in the choroidal region (i.e.,
outer to the Bruch's membrane), which arises from the absence of
the RPE, other parts of the outer retina, and possibly the
choriocapillaris. The RPE and choriocapillaris are two tissue
layers, hyperreflective in OCT scans, that normally cause the
incident light to scatter, thus partially preventing deeper
transmission of light (and therefore OCT signal) into the choroid.
OCT allows cross-sectional visualization that permits image readers
to characterize microstructural alterations in the different
laminae of the retina. Using only one type of scan for documenting
both en face and cross-sectional images of the retina, it can
therefore provide more detailed insight in retinal alterations of
GA patients than fundus autofluorescence imaging.
[0018] However, there are a number of problems with current
techniques for quantifying GA. For example, techniques that rely on
simple signal integration only indirectly address the physical
phenomenon of decreased attenuation that is actually occurring.
Similarly, techniques based on signal integration are subject to
multiple types of signal shadowing: (1) beneath retinal blood
vessel locations (though vessel sizes tend to be small in the
macula); and (2) from cataracts, in which entire areas could be
affected. Such methods also utilize spectral domain OCT (SD-OCT)
with an 800 nm wavelength. It should be noted that 800 nm light is
significantly more affected by cataracts than light with longer
wavelengths. The intensity of signal that is significantly
attenuated (e.g., 800 nm signal in the choroid or sclera) can be
artificially elevated by the OCT system's noise floor. As a
technique, this will serve to increase measurement variability and
reduce methodological sensitivity/specificity. For the sub-RPE slab
technique, the complexity of the OCT signal in the choroid can lead
to some degree of randomness in the integrated signal and resulting
analysis. The inner choroid, including the choriocapillaris and
Sattler's layer, includes many high intensity pixels, but with
great variation both in intensity and spatially. The outer choroid,
consisting of Haller's layer, comprises many pixels of lower
intensity corresponding to large blood vessels, but the size (both
in terms of width and thickness) and spacing of such vessels can
vary widely. Meanwhile, full OFI techniques (signal summation over
a full A-line) are subject to noises related to inner retinal
signals that are completely independent of AMD phenotypes. These
deficiencies can produce complexities that make it difficult for
image processing algorithms to effectively and robustly detect
atrophic regions.
[0019] In addition to the above, ANSI, and possibly other standards
organizations as well, places safety limits on the maximum power
density that can be applied to the ocular surfaces such as the
cornea and retina. This means that there is an effective limit with
respect to the achievable sensitivity for in vivo ocular imaging.
With this in mind, it is not possible to arbitrarily increase the
light power in order to achieve any desired signal intensity at
deeper retinal positions.
BRIEF SUMMARY OF THE INVENTION
[0020] Accordingly, a method of detecting geographical atrophy is
desired that overcomes the above limitations and deficiencies of
the current systems and methods.
[0021] According to one example, a method of processing acquired
ophthalmic image data comprises: providing an imaging signal
produced by an imaging device capable of penetrating beyond the
retinal pigment epithelium (RPE) of a subject's eye; detecting
intensities of the imaging signal as the imaging signal is
backscattered by each of a plurality of tissue layers in the
subject's eye for a plurality of axial scans; determining a ratio
of the intensities of the backscattered imaging signals, the ratio
being the intensity of the backscattered imaging signal of at least
a portion of a retinal layer with respect to the intensity of the
backscattered imaging signal of at least a portion of a sub-RPE
layer for each of the plurality of axial scans; and determining a
representative value for each of the plurality of axial scans based
at least in part on the determined ratio for the corresponding
axial scan.
[0022] In various embodiments of the above example, at least a
portion of a retinal layer comprises a combination of retinal
layers, and said at least a portion of a sub-RPE layer comprises a
combination of sub-RPE layers; the representative value is selected
from the group consisting of: attenuation coefficients, integrated
attenuation, or a monotonic or near-monotonic proxy measurement;
the method further comprises: identifying diseased areas of the
subject's eye based at least in part on the determined
representative values; the step of identifying diseased areas of
the subject's eye further comprises: generating a map of the
representative values or ratios, generating seeds of diseased
areas, removing outlier seeds, growing a region encompassed by the
generated seeds that were not removed, refining a contour of the
grown regions, identifying an area inside the contour as diseased,
and outputting the generated map with a contour around the regions
identified as diseased or outputting a binary mask of the regions
identified as diseased; the step of generating seeds is performed
by removing noise from the generated map and applying a
thresholding technique on the generated map; the thresholding
technique comprises finding an Otsu threshold of the generated map,
comparing the Otsu threshold with a pre-set value, and selecting an
intensity threshold based on the comparison, wherein seeds are
generated using pixels of the map that have intensities lower than
the selected intensity threshold; the step of removing outlier
seeds is performed by grouping connected seed components and
applying a distance analysis on the generated seeds; the imaging
signal has a center wavelength of at least 1 .mu.m; and/or the
retinal layer, portion of the retinal layer, combination of retinal
layers, sub-RPE layer, portion of a sub-RPE layer, or combination
of sub-RPE layers is determined using polarization sensitive
optical coherence tomography (PS-OCT).
[0023] According to another example, a method of processing
acquired ophthalmic image data comprises: providing an imaging
signal produced by an imaging device capable of penetrating beyond
the choroid/sclera interface of a subject's eye; detecting
intensities of the imaging signal as the imaging signal is
backscattered by each of a plurality of tissue layers in the
subject's eye for a plurality of axial scans; determining a ratio
of the intensities of the backscattered imaging signals for each of
the plurality of axial scans, the ratio being the intensity of a
first portion of the backscattered imaging signal with respect to
the intensity of a second portion of the backscattered imaging; and
determining a representative value of each of the plurality of
axial scans based at least in part on the determined ratio for the
corresponding axial scan.
[0024] In various embodiments of the above example, the
representative value is selected from the group consisting of:
attenuation coefficients, integrated attenuation, or a monotonic or
near-monotonic proxy measurement; the method further comprises:
identifying diseased areas of the subject's eye based at least in
part on the determined representative values; the step of
identifying diseased areas of the subject's eye further comprises:
generating a map of the representative values or ratios, generating
seeds of diseased areas, removing outlier seeds, growing a region
encompassed by the generated seeds that were not removed, refining
a contour of the grown regions, identifying an area inside the
contour as diseased, and outputting the generated map with a
contour around the regions identified as diseased or outputting a
binary mask of the regions identified as diseased; the step of
generating seeds is performed by removing noise from the generated
map and applying a thresholding technique on the generated map; the
thresholding technique comprises finding an Otsu threshold of the
generated map, comparing the Otsu threshold with a pre-set value,
and selecting an intensity threshold based on the comparison,
wherein seeds are generated using pixels of the map that have
intensities lower than the selected intensity threshold; the step
of removing outlier seeds is performed by grouping connected seed
components and applying a distance analysis on the generated seeds;
and/or the imaging signal has a center wavelength of at least 1
.mu.m or is a polarization sensitive optical coherence tomography
(PS-OCT) signal.
[0025] According to still another example, a method of processing
acquired ophthalmic image data comprises the steps of: providing an
imaging signal produced by a imaging device, the imaging signal
having a center wavelength of at least 1 .mu.m or being a
polarization-sensitive optical coherence tomography (PS-OCT)
signal; detecting intensities of the imaging signal as the imaging
signal is backscattered by each of a plurality of tissue layers in
a subject's eye for a plurality of axial scans; determining a ratio
of the intensities of the backscattered imaging signals for each of
the plurality of axial scans, the ratio being the intensity of a
first portion of the backscattered imaging signal with respect to
the intensity of a second portion of the backscattered imaging;
identifying diseased areas of the subject's eye based at least in
part on the determined representative values.
[0026] In various embodiments of the above example, the method
further comprises determining a representative value of each of the
plurality of axial scans based at least in part on the determined
ratio for the corresponding axial scan, wherein the representative
value is selected from the group consisting of: attenuation
coefficients, integrated attenuation, or a monotonic or
near-monotonic proxy measurement; the step of identifying diseased
areas of the subject's eye further comprises: generating a map of
the representative values or ratios, generating seeds of diseased
areas, removing outlier seeds, growing a region encompassed by the
generated seeds that were not removed, refining a contour of the
grown regions, identifying an area inside the contour as diseased,
and outputting the generated map with a contour around the regions
identified as diseased or outputting a binary mask of the regions
identified as diseased; the step of generating seeds is performed
by removing noise from the generated map and applying a
thresholding technique on the generated map; the thresholding
technique comprises finding an Otsu threshold of the generated map,
comparing the Otsu threshold with a pre-set value, and selecting an
intensity threshold based on the comparison, wherein seeds are
generated using pixels of the map that have intensities lower than
the selected intensity threshold; the step of removing outlier
seeds is performed by grouping connected seed components and
applying a distance analysis on the generated seeds.
BRIEF DESCRIPTION OF SEVERAL VIEWS OF THE DRAWING
[0027] FIG. 1 illustrates the anatomy of the eye as referred to
herein;
[0028] FIG. 2A illustrates one embodiment of a ratio map showing
geographic atrophy;
[0029] FIG. 2B illustrates a B-scan corresponding to a cross
section of the map of FIG. 2A through a section containing
geographic atrophy;
[0030] FIG. 3 illustrates one example of a general framework
described herein to identify, measure, and/or visualize signal
attenuation associated with atrophy;
[0031] FIG. 4 illustrates a flow chart of one example of the method
described herein to identify and/or measure signal attenuation
associated with atrophy;
[0032] FIG. 5A illustrates a flow chart of the seed generation
steps of the method of FIG. 4;
[0033] FIG. 5B illustrates the extracted seeds on the ratio
map;
[0034] FIG. 5C illustrates a flow chart of an example of a flexible
thresholding technique based on Otsu's method;
[0035] FIG. 6A illustrates a flow chart of the seed refinement
steps of the method of FIG. 4;
[0036] FIG. 6B illustrates the extracted seeds with the outliers
removed;
[0037] FIG. 7 illustrates region growing results;
[0038] FIG. 8A illustrates a contour of GA on the ratio map;
and
[0039] FIG. 8B illustrates the corresponding GA mask.
DETAILED DESCRIPTION OF THE INVENTION
[0040] Certain terminology is used herein for convenience only and
is not to be taken as a limitation on the present invention.
Relative language used herein is best understood with reference to
the drawings, in which like numerals are used to identify like or
similar items. Further, in the drawings, certain features may be
shown in somewhat schematic form.
[0041] It should be noted that while this disclosure is based on
the OCT imaging modality, it is not limited to OCT. For example, it
is applicable to ultrasound as well. It should also be noted that
the term "retina" typically refers to the tissue inner to the
retinal pigment epithelium (RPE) cell bodies. For the purposes of
this disclosure, the retina is defined as the tissue layers
including and inner to the RPE cell bodies. Therefore, the choroid
and sclera are not considered retinal tissue as used herein. A more
detailed illustration of the anatomy and relative locations of
layers of the eye and retina as referred to herein is shown in FIG.
1 on a typical OCT cross-sectional image (B-scan) 100. FIG. 1
illustrates the Inner Limiting Membrane 102, Nerve Fiber Layer
(NFL) 104, Ganglion Cell Layer (GCL) 106, Inner Plexiform Layer
(IPL) 108, Inner Nuclear Layer (INL) 110, Outer Plexiform Layer
(OPL) 112, Outer Nuclear Layer (ONL) 114, External Limiting
Membrane (ELM) 116, Retinal Pigmented Epithelium (RPE) 118, Choroid
(120), Sclera 122, and Chorioscleral Interface (CSI) 124. The
Vitreous and Foveal Pit are also shown. With further reference to
FIG. 1, the Inner Segments (IS) range roughly between the ELM 116
and the Inner Segment Ellipsoids (ISe, also known as the IS/OS
junction) 130. The Outer Segments (OS) are roughly between the ISe
and Retinal Pigment Epithelium (RPE) 118. The Bruch's Membrane (BM)
is generally estimated to be collocated with or immediately outer
to the outer edge of the RPE 118. The choriocapillaris is too small
to be imaged, but is immediately adjacent to and outer to the BM.
The Cone Outer Segment Tips (COST) is shown towards the center of
the scan between the ISe and the RPE.
[0042] Most commercial OCT equipment for retinal imaging to date
has utilized light in the 800 nm band. However, 1 .mu.m systems
provide benefits in both research and commercial applications.
Furthermore, swept-source OCT (SS-OCT) systems have an additional
advantage over SD-OCT in that the signal rolloff characteristics
are improved such that less relative loss is incurred at deeper
depths within an image. 4096-pixel CCDs in SD-OCT can exhibit
similar advantages of reduced signal rolloff characteristics. Using
these systems, the imaging capabilities of the choroid is improved.
Additionally, the outer choroid is typically visible (i.e., above
the level of system noise) even in cases of very thick choroids,
whereas traditionally this has often not been possible.
Furthermore, particularly with the 1 .mu.m wavelength (and
especially in the case of a 1 .mu.m SS-OCT system), the inner
portions of the sclera--and sometimes its entirety--can often be
successfully imaged.
[0043] Modern 1 .mu.m OCT systems can achieve even better depth
penetration (e.g., choroidal and scleral imaging capability), for
example, by: (1) optimizing the center wavelength (e.g., slightly
above 1040 nm); utilizing SS-OCT rather than SD-OCT, where SS-OCT
has an improved SNR rolloff profile; and utilizing a CCD camera
with a higher sensor count (e.g., 4096 sensors or pixels) in an
SD-OCT system. Additionally, 1 .mu.m light exhibits significantly
better penetration through cataracts than 800 nm light. An 800 nm
system is more likely to exhibit significant shadowing in A-lines
affected by cataracts. 1 .mu.m light, as it is in the infrared
range, is invisible to the patient. This reduces the degree of
stimulation to the eye during a scan, and may lead to reduced eye
motion in the captured OCT data. Therefore, the captured 3D OCT
data itself may be of higher fidelity, potentially leading to fewer
side effects in the subsequent image processing operations.
Additionally, 800 nm SD-OCT exhibits greater retinal signal
attenuation than 1 .mu.m SS-OCT. For example, this attenuation can
be about 5-10 dB at depths at and beyond the RPE.
[0044] Polarization sensitive OCT (PS-OCT) is an OCT technique that
takes advantage of changes to a light's polarization state that has
occurred in the tissue being imaged. During PS-OCT, a light source
is polarized in the reference and probe arms of the OCT system. The
backscattered light can then be detected in two orthogonal
polarization channels to detect changes in polarity between the
source light and the backscattered light. This technique is
particularly useful in retinal imaging because the retina contains
both birefringent and depolarizing tissues. Thus, PS-OCT offers
increased sensitivity for differentiating between various retinal
layers.
[0045] Furthermore, it should also be noted that while the
methodology disclosed herein is by no means limited to 1 .mu.m
and/or SS-OCT systems (and/or SD-OCT 4096-pixel CCDs) or PS-OCT,
its robustness can be enhanced by the improved imaging penetration
depth associated with such technologies. These enhancements, for
example, make possible the relatively reliable use of the
choroid/sclera interface (CSI) boundary and/or scleral signal
integrations. For example, the disclosed methodology may be
applicable to 800 nm OCT systems, including spectral domain OCT
(SD-OCT), swept-source OCT (SS-OCT), and PS-OCT.
[0046] The present disclosure is also applicable towards a scan of
any suitable dimension. For example, traditional 6 mm by 6 mm and
wide scan protocols such as 12 mm by 9 mm, as well as scans of
arbitrary dimensions, are envisioned to be within the scope of the
present disclosure. Two-dimensional (2D) and three-dimensional (3D)
scans of various dimensions are also envisioned. Accordingly, the
use of particular scan dimensions in the present disclosure is not
intended to represent limiting embodiments.
[0047] In one or more aspects described herein, systems and methods
for atrophy identification, measurement, and/or visualization of
regions of reduced tissue attenuation are based on either: (1) a
ratio of integrated signals; or (2) a ratio-less signal
integration. In one embodiment, this aspect can be used in
conjunction with 3D OCT scans for the diagnosis, monitoring, and/or
treatment of GA. However, this aspect of the invention is also
applicable to other diseases with atrophy that are detectable via
attenuation-related analysis, and via modalities other than OCT
(e.g., ultrasound).
[0048] Ultimately, the aspects described herein can be used to
determine and/or measure one or more of the following: (1) which
locations specifically are diseased; (2) the area of individual,
disease affected regions; (3) the number of individual, disease
affected regions; (4) the total area of disease affected regions;
(5) the circumference of individual disease affected regions, and
(6) the total circumference of disease affected regions. This data
can be monitored for diagnostic purposes and for monitoring the
progression of a diagnosed disease over a period of time.
Furthermore, because ratios of integrated signals themselves
objectively represent a physically meaningful phenomenon related to
attenuation, the ratios may be a useful indicator and/or predictor
of GA (or AMD in general) both in areas of complete atrophy and in
surrounding regions. Relative differences between ratios in nearby
areas might be similarly useful.
[0049] In one embodiment related to the ratio of integrated
signals, the ratio could be based on signal integration of the
retinal signal versus the choroidal and/or scleral signal for
determining GA. It should be noted that the layer combinations
could vary for other modalities and other diseases, without varying
from the scope of the present invention. In other embodiments, the
signal integration (i.e., the ratio-less) methodology could
correspond to a fixed-depth integration. For example, the inner
border of the depth to be integrated could correspond to either at
or outer to the Bruch's membrane or the distal boundary of the RPE
cell bodies when the Bruch's membrane itself cannot be resolved.
Again various layers, parts of layers, and combinations of layers
could be used without deviating from the scope of the present
invention.
[0050] The inner retinal layers are used in the ratio based aspect
and, as used herein, are defined as any of: (1) the RPE itself; (2)
the inner limiting membrane (ILM) to the Bruch's membrane
boundaries; (3) the NFL/GCL to the Bruch's membrane boundaries; (4)
the outer plexiform layer OPL/outer nuclear layer (ONL) to the
Bruch's membrane boundaries; (5) a fixed depth thick layer where
the inner border of the fixed depth to be integrated could
correspond to either at or outer to the Bruch's membrane or the
distal boundary of the RPE cell bodies when the Bruch's membrane
itself cannot be resolved; or (6) any combination of retinal
layers. The outer retinal layers are used in both the ratio and
ratio-less based aspects and, as used herein, can include all or
portions of the choroid and/or sclera. For example, this means that
the outer retinal layers could include only the inner portion of
the choroid (e.g., ignoring the Haller's layer). In some
embodiments, when using only a scleral signal, it might be
beneficial to ignore feeder vessels.
[0051] The sclera has a more uniform signal distribution than the
choroid (which has three major sublayers and is additionally highly
varied spatially due to dynamic blood vessel morphologies). This
represents a relative advantage for the sclera. However, due to
equipment sensitivity considerations, the choroid (and particularly
its inner portion) may be more practical. Segmentation accuracy of
the Bruch's membrane/choroid boundary versus the CSI is another
practical consideration.
[0052] Segmentation of intensity-only images (e.g., traditional
B-scan OCT images) can often contain errors in atrophic regions
because it is difficult for layer detection algorithms to
differentiate between healthy tissue (e.g., the RPE) and
extracellular material that might resemble the healthy tissue
(e.g., the atrophic regions). As mentioned above, the use of PS-OCT
systems can provide increased sensitivity in this regard. With
PS-OCT, healthy tissue layers can be better identified using the
derived characteristic degree of polarization uniformity (DOPU) and
a thresholding methodology. For example, the DOPU for identifying
and segmenting the RPE can be found by first calculating the Stokes
vector S on a pixel-by-pixel basis according to:
S = ( I Q U V ) = ( A x 2 + A y 2 A x 2 - A y 2 2 A x A y cos
.DELTA.O 2 A x A y sin .DELTA.O ) ##EQU00001##
Then, the DOPU may be calculated according to:
DOPU= {square root over
(Q.sub.m.sup.2+U.sub.m.sup.2+V.sub.m.sup.2)}
Finally, the RPE may then be identified by segmenting out pixels
with DOPU levels below a predetermined threshold, for example,
0.65.
[0053] For simplicity, the following disclosure describes pixel
intensity integrations as a fixed-depth integration (such that
integration occurs over a fixed number of pixels, corresponding to
a constant depth) for the outer layer calculation. However, it
should be noted that for the outer reference layer, it is
mathematically equivalent to take either the sum over a fixed depth
or to take the mean pixel intensity over either a fixed or variable
depth. Calculating over a variable depth may involve making a
modeling approximation, but the calculation results can still be
expected to convey physical meaning Averaging over a variable depth
may allow the calculation methodology to increase the degree of
averaging (i.e., include more pixels in the mean calculation),
which can serve to reduce the effects of Gaussian and/or speckle
noise. Therefore, even though it may not be ideal from a physical
equation point of view, the reduction in noise may outweigh
physical concerns. Furthermore, other calculation schemes could be
both practical and beneficial, considering the complex nature of
choroidal signal in OCT images. For example, taking the maximum
intensity projection or an average of the n maximum values within a
region of interest might in some circumstances yield more
consistent image processing results than simple integrations.
Likewise, taking the median or some arbitrary quantile of the pixel
signal intensity levels could also be performed. Similar variations
to the inner layer combination calculations can be performed. For
example, instead of integrating the signal, the mean pixel value
can be calculated. It should be noted that if the inner layer
calculation depth were to vary from A-scan to A-scan, this might
reduce the accuracy of the ratio's physical meaning, but the
calculation results could still be a useful towards identifying
regions of geographic atrophy.
[0054] In one or more embodiments, the calculation of a ratio of
signal intensities of backscattered light (one part of the ratio
representing a reference layer) is a key component of an
attenuation calculation for determining atrophy. The ratio of a
mean or integrated signal intensities calculation result could be
directly utilized as a proxy indicator for integrated attenuation
itself. Although integration over a fixed depth may better
correspond to the physical equations, the mean calculation
typically will also work well.
[0055] Rather than taking the ratio of a mean or integrated signal,
individual ratios may be found for combinations of intensities of
backscattered light from the inner retinal layers and outer retinal
layers in an A-line in order to estimate the ratio value. For
example, using a fixed-depth across all A-scans for each of the
inner and outer regions, individual ratios may be found for pixel
intensities at corresponding depths. The individual ratios may then
be combined by taking the sum, mean, median, quantile or other
similar statistical function. Another similar method envisioned
involves calculating a ratio for every combination of pixels in the
inner and outer regions. The sum of every ratio for each pixel in
the outer region can then be calculated to find a single ratio
value for each outer region pixel. Finally, the sum, mean, median,
or similar statistical function can be calculated for each of the
single ratio values, which can optionally be scaled by a desired
constant. That is, for example,
R=.SIGMA..sub.x+1.sup.d1.SIGMA..sub.n+1.sup.d2(Ti.sub.n/To.sub.x),
where x and n represent pixel locations with depths d1 and d2 for
the inner and outer regions, respectively, and Ti and To represent
the intensities of the inner region pixels and outer region pixels,
respectively, within a given A-scan. In this way, the depths
corresponding to the inner and outer regions may be different. With
small variations to the preceding equation, the depths d1 and/or d2
may even vary from A-scan to A-scan.
[0056] An attenuation coefficient in units of inverse distance
(e.g., mm.sup.-1) can be determined according to:
.mu. ATT = log ( R / .beta. + 1 ) 2 d ##EQU00002##
where .beta. is modeled as a constant value (subject to variation
somewhat by subject, location, and layer) which can be estimated
via empirical methods, R is the ratio value, .mu..sub.ATT is the
measurement value and represents a total attenuation of light over
a depth d. This coefficient can be modified to find a unitless
integrated attenuation measurement that is monotonic with respect
to the ratio value and determined according to the following
relation:
.mu. ATT d = log ( R / .beta. + 1 ) 2 .apprxeq. f ( R )
##EQU00003##
where f(R) represents a monotonic function. Although .beta. may be
subject to some variation across a volume, on a relative localized
basis monotonicity would still likely be preserved as long as such
variation is relatively gradual. In some cases, d may be
effectively kept constant by, for example, integrating over a fixed
depth or taking the mean of the integration, which is
mathematically equivalent to integrating over fixed depth except
that another constant (such as .beta.) would be made to reflect the
induced scaling factor.
[0057] In the above, any sub-RPE signal may be used as a reference
signal for determining the attenuation of the RPE. In this case,
the ratio value is defined according to:
R = T RPE T subRPE = .alpha. RPE ( 2 .mu. RPE d RPE - 1 ) .alpha.
subRPE ( 1 - 2 .mu. subRPE d subRPE ) = .beta. ( 2 .mu. RPE d RPE -
1 ) ##EQU00004## and ##EQU00004.2## .beta. = .alpha. RPE .alpha.
subRPE ( 1 - - 2 .mu. subRPE d subRPE ) ##EQU00004.3##
where T represents the total backscattered light of a layer or
combination of layers and .beta. is globally constant, given
practical computational concerns.
[0058] These equations relate the aggregated RPE signal and an
aggregated sub-RPE signal to arrive at the integrated attenuation
of the RPE. On the RPE (inner) end, other layers and combination of
layers can be used for both theoretical and practical reasons. In
GA, for example, atrophy is actually more extensive than just the
RPE, extending to the outer and inner segments, the outer nuclear
layer, and the choriocapillaris. By including any such layers, they
can be factored into the computations. In GA, atrophy can cause the
RPE to degrade and virtually disappear. It may be computationally
useful, therefore, to include additional tissue in the model with
which to calculate attenuation-related parameters. It can be
challenging to achieve highly accurate layer segmentation,
especially for diseased retina, and especially with automated
algorithms. Therefore, PS-OCT (as described above) may be used in
embodiments where such situations arise. It is even possible to
include all or most of the inner retinal layers and achieve
reasonable results.
[0059] Therefore, in another embodiment, the above ratio value
equation may be even more generalized according to:
R = T Retina T subRPE = .alpha. Retina ( 2 .mu. Retina d Retina - 1
) .alpha. subRPE ( 1 - 2 .mu. subRPE d subRPE ) = .beta. ( 2 .mu.
Retina d Retina - 1 ) ##EQU00005## and ##EQU00005.2## .beta. =
.alpha. Retina .alpha. subRPE ( 1 - - 2 .mu. subRPE d subRPE )
##EQU00005.3##
Using this generalized model along with the above unitless
integrated attenuation may be more usefully and flexibly applied
because the calculated integrated attenuation values are dependent
on one fewer variable (the inner tissue thickness). This form thus
calculates the total integrated attenuation associated with the
retina (or retinal layers or portions of retinal layers), with an
effective emphasis on that of the RPE cell bodies.
[0060] It should also be noted that either prior to or as part of
the signal intensity integrations, OCT image pixel intensity values
can be calibrated or corrected based on signal rolloff
characteristics (corresponding to light source and detection
scheme, for example) or system noise floor characteristics.
Furthermore, for embodiments related to the ratio of integrated
signals, when the inner and outer layers or combinations of layers
are not adjacent tissue, a calibration or correction factor may be
introduced to represent any attenuation associated with
intermediate layers that might not be included in the present
attenuation model.
[0061] In one or more aspects described herein, the presumed
monotonicity of the above equations enables relative comparison of
ratio values across a given patient's ratio map of the scanned
area. In other words, it enables image processing operations on 2D
ratio maps that can be created. Additionally, inter-subject
receiver-operating-characteristic (ROC) maps can be analyzed while
still remaining within the scope of this invention.
[0062] In an example based on the preceding generalized equation,
it can be inferred that those areas with very low integrated
attenuation (which can then further be correlated to attenuation
coefficients, if desired) correspond to those areas in which the
ratio is very low and correspond to probable diseased areas,
especially when such areas are clustered. Calculating the ratio
over an entire map, therefore, combined with suitable
post-processing, enables the detection of diseased regions.
[0063] The aforementioned maps are envisioned to be of attenuation
coefficients, integrated attenuation, or a proxy measurement (such
as a monotonic or near-monotonic transform) that provides
sufficient sensitivity/specificity to detect locations of retinal
atrophy. That is, in a 2D map, one axis of the map corresponds to
A-lines of a B-scan and the other axis corresponds to the B-scans.
The value at any given point represents one of the above
measurements--that is, the value of an integrated layer(s), the
determined ratio, the attenuation coefficient, the integrated
attenuation, and the like. For purposes of this description, any
such map will be referred to as a "ratio map". A sample ratio map
and 2D B-scan corresponding to a cross section of the map through a
section 200 containing GA are provided in FIGS. 2A and 2B,
respectively. The inner retinal layer corresponds to a layer,
portions of a layer, or combination of layers inner to (and
possibly including) the RPE complex. Similarly, the outer layer
corresponds to a layer, portions of a layer, or combination of
layers outer to the RPE complex (e.g., choroid and/or sclera). It
should also be noted that the ratio values may be transformed so as
to bring them more in line with attenuation outputs, for example,
dividing by an approximated constant .beta., adding 1 to that
result, and/or then taking the log transform.
[0064] In other aspects, some type of thresholding is used to
identify the atrophic disease regions in a ratio map. Such
thresholding can be of any of type, including but not limited to:
absolute levels--e.g., a ratio below a certain level always
represents a disease condition; scan relative levels--values that
are proportionately low within a scan represent a disease
condition; and local relative levels--values that are
proportionately low within a local region.
[0065] In still other aspects, image frame averaging OCT images is
used to improve SNR in OCT images, while reducing the appearance of
both Gaussian and speckle noise. Various averaging protocols,
including but not limited to: in-place averaging--in which every
B-scan is a composite image comprising an average of multiple
single-frame B-scans; three-dimensional moving averaging--where
each frame reflects a rolling average of spatially distinct frames;
and whole-volume averaging can optionally be employed as a
pre-processing operation to improve image quality.
[0066] One example of a general atrophy detection framework
according to the above aspects and embodiments, as illustrated in
FIG. 3, comprises a 2D ratio based map that is generated from a 3D
OCT volume, including several image processing steps to identify
the atrophic region contour.
[0067] FIG. 3 may apply to both 2D (e.g., single B-scan only) and
3D image embodiments. In step 300, OCT data is acquired from any
OCT or like imaging machine (frequency domain OCT, time domain OCT
etc.) that covers macular and/or optic disc regions and the data is
segmented. Segmentation results may, for example, be boundary
heights or layer region information covering a retinal and/or
choroidal region. In step 302, a ratio or integration value map is
calculated. Such a map can be any form of integration values of
certain layers, distance measurements, ratios of integration values
or distance measurements with respect to various layers, and the
like, as previously discussed. Integration can be a summation,
average, or other mathematical form within/across partial or full
A-scans. The measurement can be based on image intensities, ideally
in linear scale. The integrated or ratio value map can also be
adjusted based on local characteristics (such as normalized by
local mean and maximum values). Adjustments may include normalizing
out the noise floor and/or sensitivity roll-off (estimated
characteristic) on a row-by-row basis. Finally, step 304 comprises
visualization, which can be 1D, 2D or even 3D depending on the
desired map. An analysis in step 306 can be any image processing
steps to extract the border/contour of a region of interest from
the ratio based map/values. More detailed examples of image
processing steps that are within the scope of analysis step 306 are
shown in FIGS. 5-8 and explained in more detail below.
[0068] A more detailed example of a framework is illustrated in
FIG. 4. The framework described above with respect to FIG. 3, and
below with respect to FIG. 4, may be implemented automatically
using a processor of the imaging device or an external device
(e.g., an end user's computer). Following imaging and segmentation
400, an integration map (e.g., whereby the total intensity for each
A-line is calculated) or a ratio map (e.g., with retinal signal
versus choroidal or scleral signal for each A-line) as discussed
above is generated 402. A ratio map can enhance GA features because
the retinal signal will be weaker as a result of possible loss of
RPE and/or outer segment (OS) while the choroidal signal will be
increased with the absence of RPE.
[0069] Next, seed selection 404, as shown in more detail in FIGS.
5A and 5B, is performed on a map 500 generated in step 402. Step
502, involving map normalization, is an optional step that can be
achieved in various ways, such as histogram equalization, image
adjustment by saturating certain percentage of low and high
intensities, or normalization by a reference map. Map de-noising in
step 504 includes but is not limited to the following image
enhancement techniques: Gaussian smoothing filter, median filter,
anisotropic diffusion filter, wavelet de-noising filter,
morphological filters, subtracting background noise, order
statistic filtering, and removing blood vessel artifacts.
[0070] Map thresholding of step 506 is used to generate extracted
seeds 508, as shown in FIG. 5B (including outliers) by applying a
desired threshold to the de-noised map. When the ratio map
normalization step is bypassed, the thresholding step applies
absolute, rather than relative, thresholds. These absolute
thresholds can result in greater sensitivity and specificity than
relative thresholds based on a normalized ratio map. It has also
been found that a total or partial RPE atrophy can be accompanied
with extracellular deposits above the Bruch's Membrane, which is
commonly seen in unifocal cases. These extracellular deposits can
impact attenuation in different ways than the RPE itself. Thus, one
threshold level may not work best for every situation. Optionally,
an adaptive method (e.g., based on an Otsu's pixel clustering
method) for customizing the seeding threshold level can identify
characteristics associated with the above mentioned unifocal cases,
for example. The adaptive method can be based on various other
thresholding techniques such as entropy based threshold technique,
other thresholding techniques including using a fixed value,
statistics (such as mean, median, maximum, minimum, and the like),
and combinations thereof. An example of a flexible thresholding
technique based on Otsu's method is illustrated in FIG. 5C. In step
510, the ratio map 500 is used to obtain the Otsu threshold
(.alpha.) and statistics (.beta.), such as the mean, from the
pixels that have a lower intensity than the Otsu threshold. Next,
if the statistics .beta. are lower than a preset estimate, seeds
are generated using pixels that have intensities lower than a first
threshold .alpha.1 in step 512; if either statistics .beta. are
greater than or equal to the preset estimate, seeds are generated
using pixels that have intensities lower than a second threshold
.alpha.2 in step 514.
[0071] Next, the extracted seeds may be refined, for example by
removing outliers 406, as shown in FIGS. 6A and 6B. In step 516,
seeds 508 are grouped by connected components. Step 518 employs a
distance analysis: the absolute center locations of each component
and/or the absolute Euclidean distance from each component to the
fovea and/or the relative Euclidean distance between components are
analyzed to remove the outliers 520. The criteria to remove the
outliers can be a simple distance thresholding, a heuristic based
on the statistical analysis of the distances (such as mean and
standard deviation) combined with the pixel numbers of the
components, cluster analysis, or various other similar techniques.
Again, the extracted seeds are shown on the ratio map in FIG. 6B,
however, the outliers have been removed.
[0072] Next, the de-noised map 504 and seeds with outliers removed
406 are used to grow the seed region in step 408. The region
growing step 408 can be performed separately on each connected
component seed or just the seeds from one component. The mean,
standard deviation, gradient information, and/or other statistical
descriptors may be used in a single iteration or multiple
iterations to grow the seed region. Region growing 408 may also
include post-processing such as hole-filling or convex hull
operations. Results of such techniques are shown in FIG. 7,
illustrating that the area of the seeds with respect to FIGS. 5B
and 6B has increased.
[0073] Finally, the results of post processing 410 and contouring
412 are shown in more detail in FIGS. 8A and 8B. During the
contouring step 412, a contour around the grown seeds is provided
and may be output as a binary mask or in a similar
visualization--that is, the area contained within the contour may
be specially designated (e.g., white) to indicate a region of
geographic atrophy, while the remaining area of the image is
designated, for example, black. Such a mask is illustrated in FIG.
8B while the contour is illustrated on the original cross-sectional
image in FIG. 8A. A contour may be refined by smoothing (e.g.,
using a moving average) or re-detecting contours for each component
with guidance from the original contour. Such guidance could
include using a graph search to re-search the contour based on the
image contrast. The overall image processing and analysis is not
limited to the region growing method. Any segmentation method, such
as active contour based segmentation or watershed segmentation,
that creates similar results as shown herein are envisioned within
the scope of the present disclosure.
[0074] While the above methods were described as automated
procedures (e.g., using a processor), it should be noted that full
automation of the methods described herein is not intended to be
limiting. Rather, in some embodiments, the methods may be entirely
manual. Semi-automated embodiments are also envisioned to be within
the scope of the present disclosure. For example, one
semi-automated embodiment may additionally include a manual
correction step(s).
[0075] The various embodiments described herein refer to imaging
data of an eye obtained from OCT systems. However, the disclosed
techniques and processes may equally apply to imaging data obtained
using other types of imaging devices, for example ultrasound.
Additionally, the disclosed techniques and processes may equally
apply to any biological tissues, in addition to those in the eye,
for which meaningful layers may be defined, segmented, and imaged
as described herein.
[0076] It is to be noted that the above aspects, embodiments, and
examples are envisioned to be implemented automatically by a
processor. A "processor" as used herein refers to any, or part of
any, electrical circuit comprised of any number of electrical
components, including, for example, resistors, transistors,
capacitors, inductors, and the like. The circuit may be of any
form, including, for example, an integrated circuit, a set of
integrated circuits, a microcontroller, a microprocessor, a
collection of discrete electronic components on a printed circuit
board (PCB) or the like. The processor may also stand alone or be
part of a computer used for operations other than processing image
data. It should be noted that the above description is
non-limiting, and the examples are but only a few of many possible
processors envisioned.
* * * * *